Tumor Heterogeneity—A ‘Contemporary Concept’ Founded on Historical Insights and Predictions

Visit the Cancer Research 75^th Anniversary timeline.

In this time of easy access to articles via electronic means, too often some older literature is overlooked or, even worse, ignored because newer articles relegate older literature to sometimes extended sessions of scrolling through pages on PubMed. In the 75^th Anniversary of Cancer Research, we take a look back at 48 articles that have had high impact on cancer research. In 1984, Gloria Heppner published Tumor Heterogeneity (1), a review in which the long-standing recognition that cancers are composed of multiple subpopulations was analyzed. This article was more often highlighted by editors, AACR Fellows, and cancer researchers than any other. Why do I think that is the case? I am struck in re-reading this article by the clarity of thought, objective presentation, and interpretation of the data, and, as importantly, by the insights that have withstood the tests of time.

All tumors (not necessarily neoplasms) are heterogeneous (2, 3). Yet, surprisingly, recent deep sequencing reports seem to hail the existence of heterogeneity as a new concept. Hardly! The literature dates back to the 1800s. Virtually every phenotype in cancer is heterogeneous. And, yes, although some of the recent deep sequencing papers elegantly verify that there is genetic heterogeneity at the single cell level (4–7), the findings parallel what was described decades ago in other systems.

What still remains 30 years hence is a still unclear understanding of the origins of tumor heterogeneity. Some might argue that the cancer “stem cell” theory explains the origins of heterogeneity (8–11); others might argue that genomic/genetic instability drives diversification of cancers (2, 6, 9, 12–14). In truth, both are likely drivers (Fig. 1).

Discussion of the heterogeneity in tumors is too often limited to discussions of clonal differences. But, there are, in fact, three other types of tumor heterogeneity—population, positional, and temporal (Fig. 1). Population heterogeneity refers to the uniqueness of tumors arising from the same cell type in different patients. That is, each patient's tumor is unique, albeit sharing some properties. Underlying all of the oncogenic and genetic instability drivers are quantitative trait loci that control cellular properties (15, 16) that contribute to interpatient variability.

Likewise, patient-specific selective pressures modulate the progression of individual cancers. Such are the parameters (e.g., hormonal or immune status, stress, nutrition, therapies) that contribute to temporal heterogeneity. A biopsy is merely a snapshot of a small subset of neoplastic cells at a given moment in time. Tumor progression is an evolutionary process in which tumor cells with a selective advantage begin to appear as clonally dominant, whereas disadvantaged cells are selected against (17–19). Thus, at any given time, tumor composition is unique.

Likewise, cellular phenotypes are influenced by the signals they receive, e.g., pO₂, growth factor concentration, matrix biophysical properties, and signals from other tumor cells. For example, disseminating cells that have been hypoxic are more likely to successfully form macroscopic metastases than cells that have been adequately oxygenated (20–22). Similarly, responses to therapy can be influenced by proximity to vessels, e.g., cells nearer to blood vessels are more oxygenated and therefore sensitive to ionizing radiation, or cells distant from vessels may receive lower drug concentrations because of the diffusion distance (23).

These considerations were eloquently addressed by Dr. Heppner in her review. She ultimately showed the similarities between tumor progression, heterogeneity development, and evolutionary theory. Those parallels profoundly shaped a generation of researchers' and clinicians' thinking about how to address heterogeneity in experimental design and treatment. Tumor cell societies, in which the distinct subpopulations of cancer cells influence the behavior of other cells in the mass (or even at distant sites in the body; Fig. 1), are now well appreciated, but were first introduced in her review.

Although the above insights have proven helpful, an even deeper understanding about heterogeneity is needed, especially in light of new research models and the emergence of personalized treatment strategies. In recent years, some have argued that cell lines are not as heterogeneous as in situ tumor implants or xenografts. However, subcloning argues against that point (reviewed in ref. 2). Certainly, the vast majority of tumor cells are genetically unstable both in vitro and in vivo (2, 13). Some assume that cell clones remain homogeneous, but the data overwhelmingly argue against that assumption (3, 13). Still others argue that patient-derived xenografts (PDX) are more reflective of their state in patients; however, unnatural selective pressures associated with growing them in non-human hosts (i.e., with murine vessels providing nutrients) coupled with genetic instability that still exists (24) certainly portends continued evolution. So, although there are certainly some advantages to PDX models, there will be artificialities associated with their use, just as there are with all experimental models. Those limitations will emerge as PDX models are more widely utilized.

So, although new ways of defining the heterogeneity that exists within a tumor mass have been added since 1984, the principles outlined in Dr. Heppner's seminal review have withstood the tests of time. Reflecting upon the insights she presented continues to provide a context upon which cancer researchers should interpret their data. They further provide cautionary notes for those treating cancer—monotherapies will rarely be effective against tumors composed of mixed populations, and tumors are continually evolving ways to circumvent attempts to limit them. Oncologists battle a wily enemy.

Tumor heterogeneity exists. Heterogeneity is a hallmark of all cancers. Cancer cells do not exist in isolation—communications between tumor cells and between tumor and host cells coupled with selection pressures imposed upon the tumor combine to influence neoplastic evolution. Cancer researchers and oncologists ignore heterogeneity at their own peril.

No potential conflicts of interest were disclosed.

The author is thankful for the generous support from the National Cancer Institute, National Foundation for Cancer Research, Susan G. Komen for the Cure, and many other agencies over the years. The author apologizes to the many authors whose work was not cited due to space limitations.